proposal for capabilities lookup

Dan Libby danda at videntity.org
Thu Nov 17 15:35:26 PST 2005


On Thursday 17 November 2005 17:14, Ernst Johannes wrote:
> On Nov 17, 2005, at 14:41, Dan Libby wrote:
>
> > 9.2 OPTIONS
> >
> >    The OPTIONS method represents a request for information about the
> >    communication options available on the request/response chain
> >    identified by the Request-URI. This method allows the client to
> >    determine the options and/or requirements associated with a
> > resource,
> >    or the capabilities of a server, without implying a resource action
> >    or initiating a resource retrieval.
>
> That doesn't alleviate your concern about too many round-trips,
> however, does it? Maybe I misunderstand what you are saying.

Well, if we make a single OPTIONS request, and that request returns a YADIS 
capabilities document, then:
 a) we have one request/response instead of 2
 b) we do not have to parse any nasty and possibly huge and bandwidth wasting 
html.

>
> >    Responses to this method are not cacheable.
>
> We don't mean this one, do we? I consider capability documents to be
> eminently cacheable -- probably more so than regular content documents.

I think we could safely ignore that at the consumer.  It is a concern for 
proxy-caches. Or there may be another solution, such as specifying a max 
cache lifetime within the capabilities doc, and making that part of YADIS, 
rather than HTTP.

>
> This last part does not seem to apply.
>
> This command seems to be more of a "tell me how smart the server is"
> kind of thing than a "tell me what this resource, hosted by some
> server, can do".

I don't interpret it that way.  Specifically:

 If the Request-URI is not an asterisk, the OPTIONS request applies
 only to the options that are available when communicating with that
 resource.

Thus, the returned document should specify capabilities for the requested 
resource ( URL ). 

The biggest issue seems to be that the spec is silent about the format of the 
returned document.  But I think that is fine so long as the consumer sends an 
Accept header that requests a yadis doc.


Anyway, I raise it as a possibility.

-- 
Dan Libby

http://videntity.org/
 - One identity to login with them all
 - Social networking between websites and blogs


More information about the yadis mailing list